Measured vibration signals from rolling element bearings with defects are generally nonstationary, and are multiscale in nature owing to contributions from events with different localization in time and frequency. This paper presents a novel approach to characterize the multiscale signature via empirical mode decomposition (EMD) for rolling bearing localized defect evaluation. Vibration signal measured from a rolling element bearing is first adaptively decomposed by the EMD to achieve a series of usable intrinsic mode functions (IMFs) carrying the bearing health information at multiple scales. Then the localized defect-induced IMF is selected from all the IMFs based on a variance regression approach. The multiscale signature, called multiscale slope feature, is finally estimated from the regression line fitted over logarithmic variances of the IMFs excluding the defect IMF. The presented feature reveals the pattern of energy transfer among multiple scales due to localized defects, representing an inherent self-similar signature of the bearing health information that is embedded on multiple analyzed scales. Experimental results verify the performance of the proposed multiscale feature, and further discussions are provided.
Rolling Bearing Localized Defect Evaluation by Multiscale Signature via Empirical Mode Decomposition
Contributed by the Design Engineering Division of ASME for publication in the JOURNAL OF VIBRATION AND ACOUSTICS. Manuscript received June 9, 2011; final manuscript received April 5, 2012; published online October 29, 2012. Assoc. Editor: Alan Palazzolo.
- Views Icon Views
- Share Icon Share
- Search Site
He, Q., Li, P., and Kong, F. (October 29, 2012). "Rolling Bearing Localized Defect Evaluation by Multiscale Signature via Empirical Mode Decomposition." ASME. J. Vib. Acoust. December 2012; 134(6): 061013. https://doi.org/10.1115/1.4006754
Download citation file: